[ewg] FW: MOIs for FDR devices RX tolerance testing - our proposals and questions

Wed Mar 13 09:27:04 PDT 2013

Here is Brent's comment regarding the proposed Rx Tolerance test.

From: Rothermel, Brent R [mailto:brent.r.rothermel at intel.com] 
Sent: Wednesday, March 13, 2013 8:37 AM
To: Rupert Dance; Cole, Cliff
Cc: bennera at us.ibm.com
Subject: RE: MOIs for FDR devices RX tolerance testing - our proposals and
questions

Rupert,

I agree with some of their proposals in this email, not with others

Agree

.         The DUT transmitter should be used to generate the
counter-propagating aggressors.  This DUT transmitter should have a
'compliant' edge rate, meaning no faster than 24ps but fast enough to meet
the eye mask.

.         A cable can be used as the stimulus *if* it has worst-case
compliant signal coming out of it.  This is unlikely to be the case, which
is why we have used a pattern generator to create this signal in the past

Disagree

.         We cannot go through both an MCB and an HCB to get into the system
under test - this is too much loss.

.         The 'crosstalk generator' in their diagram is only used for
calibration of the victim signal, and should be removed when connected to
the system under test (because the system under test is generating the
counter-propagating crosstalk at that point.

.         The co-propagating aggressors should be sourced from the signal
generator and propagating through the cable into the system under test.

I look forward to discussing this with Mellanox on today's EWG meeting.

Thanks

Brent Rothermel

Intel Corporation

(610) 312-2171 

8-322-2171

brent.r.rothermel at intel.com 

From: Rupert Dance [mailto:rsdance at soft-forge.com] 
Sent: Monday, March 11, 2013 9:03 PM
To: Rothermel, Brent R; Cole, Cliff
Cc: bennera at us.ibm.com
Subject: FW: MOIs for FDR devices RX tolerance testing - our proposals and
questions

Hi Guys,

I am trying to spread the load of these questions around the group but we
did not get time to review this proposal at the last EWG meeting. Mellanox
is proposing a RX tolerance testing for use with limiting active cables and
they have several questions and here is the summary of the question I would
like you to consider:

  _____  

The FDR specification requires the signals, transferred over the
counter-propagating aggressor lines, to have a 700 mV amplitude with 24 ps
transition time, as measured at TP6a (Figure 179 in the FDR specification).
As it was mentioned, these signals are generated by the transmitters of the
tested FDR port, and therefore should have a minimal transition time of 24
ps (per Table 49 of the FDR specification - FDR host output specifications
at preset 0, for Limiting Active Cables). In this relation, we understand
that the specified test value for the crosstalk calibration transition time
of 24 ps is also a minimal value, otherwise it would be nearly impossible to
satisfy the test requirements for both the driver and the receiver testing.
Specifically, Mellanox transmitters provide signals with larger transition
times. Please advise whether our understanding of the specification is
correct and signals with 700 mV amplitude and a slower than 24 ps transition
time, generated by the transmitters of the tested port satisfy the
requirements for the counter-propagating aggressors.

  _____  

Thanks

Rupert

From: Alexander Rysin [mailto:arysin at mellanox.com] 
Sent: Tuesday, March 05, 2013 9:45 AM
To: Rupert Dance
Cc: Daniel Fradkin; Vitaly Roitberg; Aviv Berg
Subject: MOIs for FDR devices RX tolerance testing - our proposals and
questions

Hi Rupert,

Please find attached our proposed MOI for FDR devices RX tolerance testing
for use with linear cables for Tektronix BERTScope Bit Error Rate Tester,
for your revision. We are looking forward to any comments regarding required
changes in order to make it a useful document for the IBTA.

I have also begun working on a similar document for FDR devices RX tolerance
testing for use with limiting active cables. I have studied the FDR
specification and the related parts of the IEEE 802.3 document, which
propose the following test setup (Figure 179 in the FDR specification):

The goal of the setup is to provide a signal with a certain level of
distortion to the tested receiver. The IEEE 802.3 document has quite
specific requirements for the stressed signal. With that, it states that any
approach that creates the appropriate stress is acceptable (page 392), hence
using the exact setup is not mandatory. Given this statement, since we do
not have the time to implement the exact setup before PF23, we have searched
for an alternative setup. After some research we performed, we have found
that the appropriate stress can be achieved by employing an actual active
limiting cable (specifically, we have achieved it with a cable with P/N
018-1651-001). We feel that such approach to the test should be preferred,
since it better emulates the real situation, when the receiver is connected
to an active cable. Therefore, we propose the following setup for this test:

The limiting cable creates the required stress for the tested receiver. This
is verified by connecting an oscilloscope to the right MCB and measuring the
DDPWS, the J2 jitter and the J9 jitter. In addition, sinusoidal jitter with
amplitude of 0.05 UI and frequency higher than 20 MHz should be applied to
comply with the IEEE 802.3 requirements.

In addition to the DDPWS and jitter requirements to the stressed signal, the
FDR specification presents requirements for the crosstalk from near end
(counter-propagating) and far end (co-propagating) aggressors. We propose to
create this noise in the following manner: the signals for the
co-propagating aggressors should be generated from an external signal
generator, satisfying the requirements for these signals (450 mV amplitude
with 17 ps 20%-80% transition time, e.g. generated by Anritsu MU181020B PPG)
and the signals for the counter-propagating aggressors should naturally be
generated from the transmitters of the tested FDR port.

We would like to receive your comments regarding this approach in order to
complete the MOI document for FDR RX tolerance testing for use with limiting
cables and in order to employ it in the RX tolerance testing in the
framework of PF23. In addition, we have a question regarding the
requirements for the crosstalk from the counter-propagating aggressors. The
FDR specification requires the signals, transferred over the
counter-propagating aggressor lines, to have a 700 mV amplitude with 24 ps
transition time, as measured at TP6a (Figure 179 in the FDR specification).
As it was mentioned, these signals are generated by the transmitters of the
tested FDR port, and therefore should have a minimal transition time of 24
ps (per Table 49 of the FDR specification - FDR host output specifications
at preset 0, for Limiting Active Cables). In this relation, we understand
that the specified test value for the crosstalk calibration transition time
of 24 ps is also a minimal value, otherwise it would be nearly impossible to
satisfy the test requirements for both the driver and the receiver testing.
Specifically, Mellanox transmitters provide signals with larger transition
times. Please advise whether our understanding of the specification is
correct and signals with 700 mV amplitude and a slower than 24 ps transition
time, generated by the transmitters of the tested port satisfy the
requirements for the counter-propagating aggressors.

We have one more question regarding the PF23. To implement our proposed
setup for the RX tolerance testing. In case signals with fast transition
times (17 ps) need to be generated for the co-propagating aggressors, a
multy-channel signal generator is requred during the entire time of the
test. Can the Anritsu 14G PPG be at our disposal during PF23?

Thank You in Advance and Best Regards,

Description: Description: Description: Description: Description:
Description: cid:image001.png at 01CC8CED.AB5E8D20     Alexander Rysin

Sr. Circuit Compliancy & Characterization Engineer | Mellanox Technologies
Ltd.

Office: +972-74-7237087| Cell: +972-54-4580159| Yokneam, Israel 

Attachment saved to
file:///C:\Users\rsdance\Documents\Lamprey\CIWG\Plugfest\2013-04\Documentati
on\Device Technical Documents\Rx Tolerance Tests\icn.m
<file:///C:\Users\rsdance\Documents\Lamprey\CIWG\Plugfest\2013-04\Documentat
ion\Device%20Technical%20Documents\Rx%20Tolerance%20Tests\icn.m> 

Attachment saved to
file:///C:\Users\rsdance\Documents\Lamprey\CIWG\Plugfest\2013-04\Documentati
on\Device Technical Documents\Rx Tolerance
Tests\BERTScope_FDR_Device_RX_MOI_v1.00.pdf
<file:///C:\Users\rsdance\Documents\Lamprey\CIWG\Plugfest\2013-04\Documentat
ion\Device%20Technical%20Documents\Rx%20Tolerance%20Tests\BERTScope_FDR_Devi
ce_RX_MOI_v1.00.pdf> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openfabrics.org/pipermail/ewg/attachments/20130313/03aeefd1/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image002.png
Type: image/png
Size: 58289 bytes
Desc: not available
URL: <http://lists.openfabrics.org/pipermail/ewg/attachments/20130313/03aeefd1/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image003.png
Type: image/png
Size: 47148 bytes
Desc: not available
URL: <http://lists.openfabrics.org/pipermail/ewg/attachments/20130313/03aeefd1/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image004.png
Type: image/png
Size: 6535 bytes
Desc: not available
URL: <http://lists.openfabrics.org/pipermail/ewg/attachments/20130313/03aeefd1/attachment-0002.png>